To determine independent predictors of lymphedema (LE) after breast radiotherapy and to quantify added risks of LE from regional node irradiation (RNI).

A total of 2,579 women with T1-2, N 0-3, M0 breast cancer treated with breast conservation between 1970 and 2005 were studied. A total of 2,169 patients (84%) received radiation to the breast (B), 226 (8.8%) to the breast and supraclavicular LNs (B+SC), and 184 (7.1%) to the breast, supraclavicular LNs, and a posterior axillary boost (B+SC+PAB). Median follow-up was 81 months (range, 3-271).

Eighteen percent of patients developed LE. LE risks were as follows: 16% (B), 23% (B+SC), and 31% (B+SC+PAB) (p < 0.0001). LE severity was greater in patients who had RNI (p = 0.0002). On multivariate analysis, RT field (p < 0.0001), obesity index (p = 0.0157), systemic therapy (p = 0.0013), and number of LNs dissected (p < 0.0001) independently predicted for LE. In N1 patients, the addition of a SC to tangents (p < 0.0001) and the addition of a PAB to tangents (p = 0.0017) conferred greater risks of LE, but adding a PAB to B+SC RT did not (p = 0.8002). In the N2 patients, adding a PAB increased the risk of LE 4.5-fold over B+SC RT (p = 0.0011).

LE predictors included number of LNs dissected, RNI, obesity index, and systemic therapy. LE risk increased when a SC or PAB were added in the N1 subgroup. In the N2 patients, a PAB increased the risk over B+SC. The decision to boost the axilla must be weighed against the increased risk of LE that it imposes.